|Title:||Human face super-resolution|
|Subject:||High resolution imaging.|
Image processing -- Digital techniques.
Image processing -- Mathematical models.
Hong Kong Polytechnic University -- Dissertations
|Department:||Department of Electronic and Information Engineering|
|Pages:||98 p. : ill. ; 30 cm.|
|Abstract:||In this thesis, we have investigated different algorithms for human face super-resolution (SR), which are important for applications such as face recognition, video surveillance and application of many digital devices etc. With these face SR algorithms, face-image resolution can be increased while the facial-image quality is maintained. We have studied two types of SR algorithms: reconstruction-based and learning-based methods. For reconstruction-based methods, we have investigated and implemented the "bilinear" method and the "bicubic" method. These methods are simple, but can achieve only a limited performance, since limitation of information provided. In order to achieve a better performance, learning-based methods are usually employed; these learn the relations between low-resolution (LR) and high-resolution (HR) images from a dataset containing pairs of LR-HR pairs. We have investigated and implemented the "eigentransformation" method, which use principal component analysis (PCA) to represent a face image as a linear combination of training samples. We have proposed two improvements to this method. The first improvement is that, instead of considering the linear relations between a LR face image and the LR training samples, LR images are first super-resolved using a reconstruction-based method, and then the linear relations are computed. The other improvement is to use a face-recognition method to search similar faces to an input LR face before eigentransformation is applied. We also compare the eigentransformation methods to a patch-based method, namely position patch. We evaluate the respective performances of the different algorithms in terms of visual quality and some other objective measurements.|
|Rights:||All rights reserved|
Files in This Item:
|b25075494.pdf||For All Users (off-campus access for PolyU Staff & Students only)||3.34 MB||Adobe PDF||View/Open|
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item: